Sakatani et al. 1 Supporting Online Material Materials and methods Mice and genotyping: H19 mutant mice with C57BL/6J background carrying a deletion in the structural H19 gene (3 kb) and 10 kb of 5 flanking sequence were obtained from Shirley M. Tilghman (ref. 7). Paternal H19 heterozygotes were maintained without LOI phenotype by breeding female wild-type C57BL/6J and male H19 +/. Experimental crosses were performed between female H19 +/ and male Apc +/Min (C57BL/6J). Mice were genotyped as follows using DNA extracted from the tails with DNeasy Tissue Kit (Qiagen, Valencia, CA). For H19, PCR was performed using two forward primers and one common reverse primer to obtain a 847-bp product for wild type allele and a 1,000-bp product for mutant allele. Primer sequences and annealing temperatures were: H19-F, TCC CCT CGC CTA GTC TGG AAG CA; Mutant-F, GAA CTG TTC GCC AGG CTC AAG; Common-R, ACA GCA GAC AGC AAG GGG AGG GT; 66 C. For Apc, PCR and direct sequencing were performed using the following primers: Apc-F, TTT TGA CGC CAA TCG ACA T; Apc-R, GGA ACT CGG TGG TAG AAG CA; 55 C. Mice were sacrificed at 42 days and 120 days for tumor quantitation, histology, and immunostaining, and the entire intestine and other organs were collected. In addition, 150 day old H19 mutant mice carrying knock-in alleles of sequence change from GTGG to ATAT in three of the four CTCF target sites within H19 imprinting control region were established previously and crossed with SD7 mice as described (ref. 23). We compared paternally transmitted mutant alleles (non-loi) to maternally transmitted alleles (LOI) with immunostaining performed on the same slide. All the animal experiments were performed in accordance with University guidelines. Tumor analysis and immunostaining: For analysis of numbers and sizes of tumors, the entire intestine was flushed with cold PBS and was opened longitudinally.
Sakatani et al. 2 One half was frozen for further molecular analysis. The other half was fixed with 10% formalin and stained with 0.03% methylene blue, and numbers and sizes of tumors were measured under light microscopy, blinded for genotype. For histopathological analysis, the entire intestine and other organs were fixed in 4% paraformaldehyde followed by 70% ethanol, and embedded in paraffin. H&E staining and immunohistochemistry against musashi1 (Chemicon, AB5977, 1:200 dilution), twist (Santa Cruz Biotechnology, SC-15393, 1:100 dilution), villin (Chemicon, MAB1671, 1:100 dilution), ephrin-b1 (R&D Systems, AF473, 1:25 dilution) and PCNA (Transduction, P56720 1:200 dilution) were performed comparing 4 mice in each group over the entire length of the small intestine, to analyze basic morphology, the balance of undifferentiated to differentiated compartments, the proliferation index, and the distribution of proliferative cells. Crypt length was measured from the base of intestinal crypts to the base of intestinal villi. Determinations of crypt length were blinded to genotype and based on a minimum of 5 individual measurements of random, well oriented sections of intestine on each of 2 different histologic sections (10 sections apart), defined as an area with a minimum of three adjacent villi and associated crypts cut perpendicularly to the long axis of the bowel lumen. Mushashi1 positive cells were counted using a hemocytometer in 10 individual crypts per mouse that were perpendicularly oriented to the long axis of the intestine. Quantitative image analysis of PCNA labeling was performed using the ACIS II automated image analysis system (Chromavision, San Juan Capistrano, CA) with measurements of both the percent and intensity of positive labeling cells determined in 10 individual crypts per mouse that were perpendicularly oriented to the long axis of the intestine. The distribution of proliferative cells was determined using a modification of the method described by Lipkin et al. (ref. 14) using a hemocytometer to measure the height of the highest PCNA positive cell
Sakatani et al. 3 within an intestinal crypt divided by the overall height of that same crypt, again among 10 individual crypts per mouse. Measurements were expressed as a ratio, and the mean ratio for LOI(+) and LOI( ) mice was determined. For determinations of apoptotic rate, sections of the small intestine were evaluated for the number of positive labelling cells within a total of 20 intestinal crypts per mouse using a TUNEL Apoptotic Detection Kit (Upstate, Lake Placid, NY). The normal colonic mucosa of colonoscopy clinic patients with and without LOI, as determined in our previous study (ref. 2), were analyzed with immunostaining of musashi1 and twist. musashi1 and twist immunolabeling was evaluated independently and blindly within the bottom half of intestinal crypts, the upper half of intestinal crypts and surface epithelium. Positive labeling was scored as nuclear staining with or without cytoplasmic staining in epithelial cells. RNA and protein analysis: Total RNA was extracted from tumor and non-tumor regions of the frozen intestine using RNeasy Kit with DNase I treatment (Qiagen), and reverse-transcribed using SuperScript II (Invitrogen, Carlsbad, CA). Expression level of Igf2 was quantified by real-time RT-PCR using SYBR Green PCR Core Reagents and ABI Prism 7700 Sequence Detection System (Applied Biosystems, San Jose, CA), and normalized to that of -actin. Primers and annealing temperatures are as follows. Igf2: CAT CGT GGA AGA GTG CTG CT and GGG TAT CTG GGG AAG TCG T, 60 C. actin: TAC CAC CAT GTA CCC AGG CA and GGA GGA GCA ATG ATC TTG AT, 60 C. Homogenized samples of small intestine of 42 day mice were applied to SDSpolyacrylamide gel (16%) electrophoresis with NuPAGE LDS buffer (Invitrogen) after acidification in 1M acetic acid and lyophilization. Gels were transferred onto Immune- Blot PVDF membrane (BioRad, Hercules, CA), and the membranes were blocked with
Sakatani et al. 4 blocking buffer (5% non-fat dried milk, 0.1% Tween-20 in TBS) at 4 C overnight, incubated with a 1:500 dilution of Igf2 antibody (Upstate, Lake Placid, NY) or a 1:1000 dilution of Igf2 antibody (Abcam, Cambridge, MA) at room temperature for 1 h. After treatment with HRP conjugated secondary antibody and ECL detection reagents (Amersham, Piscataway, NJ), and exposure to X-ray film, signal intensities were measured with a scanning densitometer. The gels were stained with SimplyBlue SafeStain (Invitrogen), and the intensities of the staining were measured with a scanning densitometer to correct the signal intensities.
Sakatani et al. Fig. S1 5 A Maternal allele (WT) H19 Deletion Model Igf2 DMR H19 Maternal allele (H19 delet.) Igf2 13 kb neo x Genotypes LOI Tumor H19 +/ x Apc +/Min H19 +/+ Apc +/Min H19 /+ Apc +/Min H19 +/+ Apc +/+ + + + H19 /+ Apc +/+ + B Maternal allele (WT) H19 DMR Mutation Model Igf2 DMR H19 Maternal allele (DMR mut.) Igf2 DMR H19 x LOI SD7 x 142* 142* x SD7 + C Experiments Model Age Tumor Histology Deletion 42 days 120 days + + + + Mutation 150 days n.d. +
Sakatani et al. Fig. S2 6 A Relative Igf2 Expression 3 2 1 0 N N N N N N T T LOI + + + + Apc +/+ +/+ +/+ +/+ +/Min +/Min +/Min +/Min 42 day 120 day B Upstate 05-511 Abcam ab9574 kda 30 LOI( ) LOI(+) LOI( ) LOI(+) 20 14 6.5
Sakatani et al. Fig. S3 7 A B
Sakatani et al. Fig. S4 8
Sakatani et al. Fig. S5 9 A B C D E F
Sakatani et al. Fig. S6 10
Sakatani et al. Fig. S7 11
Sakatani et al. Fig. S8 12 A B *
Sakatani et al. 13 Figure legends for supplementary figures Fig. S1. Mouse models of H19 deletion and DMR mutation. (A) H19 deletion model (ref. 7). Thirteen kb including the H19 gene and its DMR in the upstream region were replaced with neo. When this deletion is inherited from the mother, H19 expression is lost and the normally silent Igf2 allele is activated as shown. Experimental crosses were performed between female H19 +/ and male Apc +/Min mice to obtain the four genotypes shown. (B) H19 DMR mutation model (ref. 23). Three of the four CTCF binding cites at H19 DMR were mutated (closed boxes). When this mutation is inherited from the mother, the normally silent Igf2 allele is activated with H19 expression maintained (see also Fig. S6 and S7). DMR-mutant (142*) female or male mice were crossed with wild type SD7 to obtain mice with LOI and normal imprinting of Igf2, respectively. (C) Experiments performed with each model. Fig. S2. Igf2 mrna and protein levels. (A) Relative Igf2 mrna level. Igf2 mrna levels were analyzed by real-time RT-PCR, normalized to that of -actin, and are displayed relative to the small intestine of wild type LOI( ) mice at 42 days. Igf2 mrna was 2.0-fold greater in the non-tumor region of LOI(+) mouse intestine than in LOI( ) mouse intestine at 42 days (P=0.002), and 2.1-fold greater at 120 days (P=0.04). For LOI(+) Min mice at 120 days, Igf2 mrna showed a 2.2-fold increase in the non-tumor region (P=0.03) and a 2.3-fold increase in the tumor region (P=0.003), compared with LOI( ) Min mice. Within a given genotype, the expression of Igf2 did not increase from normal to tumor, consistent with an early role for LOI in tumorigenesis. N, non-tumor region. T, tumor region. P values were calculated by T-test for each comparison. (B) Western blot analysis of Igf2 protein. Signals were detected at 15 kda, 17 kda and weakly at 18 kda using two separate antibodies (shown), and the intensities were
Sakatani et al. 14 increased 1.7-2.1 fold (Upstate) and 1.5-2.1 fold (Abcam) in the small intestine of LOI(+) mice, normalized to total protein. These higher molecular weight forms are well described in mammals (S1-S4) and are more efficient activators of the Igf1 receptor (the signaling target of Igf2) than is the fully processed form of Igf2 (S2). Fig. S3. Histomorphology of small intestinal mucosa in LOI( ) mice (A) versus LOI(+) mice (B). Detailed histopathological exam of the small intestine, colon, and extraintestinal tissues were performed in both 42 day and 120 day (shown) mice. Although no architectural differences are seen in association with LOI status, the crypt length of the small intestine of LOI(+) mice showed a statistically significant increase compared to their wild-type littermates: 1.2-fold increase (15.3 ± 1.9 m vs. 13.1 ± 1.8 m, P<0.01) at 42 days; and 1.5-fold increase (19.6 ± 2.0 m vs. 13.0 ± 2.0 m, P<0.0001) at 120 days. Scale bars, 10 m. Fig. S4. Immunohistochemistry for villin and ephrin-b1 in 42 day mice. (A) In LOI( ) mice, villin is found in a cytoplasmic distribution throughout differentiated enterocytes lining intestinal villi, with expression extending to the transition zone and superficial crypts. (B) In LOI(+) mice, villin is largely restricted to the enterocytes lining intestinal villi with no expression noted within the transition zone or superficial crypts (indicated by arrow), consistent with a contraction of the differentiated cell compartment. (C) ephrin-b1 protein expression shows a similar pattern as described for villin, seen as cytoplasmic labeling of differentiated enterocytes lining intestinal villi in LOI( ) mice. (D) In contrast, immunostaining for ephrin-b1 is markedly decreased in the intestinal villi of LOI(+) mice. Scale bars, 10µm.
Sakatani et al. 15 Fig. S5. Immunohistochemistry for musashi1 and twist in 42 day mice. (A) In LOI( ) mice, musashi1 expression is detected within the cytoplasm and nuclei in rare cells within intestinal crypts (representatively indicated by arrow). (B) In LOI(+) mice, musashi1 cytoplasmic and nuclear labeling is detected throughout the intestinal crypts. (C) In LOI( ) mice, no musashi1 expression is detected within the overlying intestinal villi. (D) In LOI(+) mice, ectopic cytoplasmic and nuclear expression is seen in enterocytes lining intestinal villi. (E) Weak cytoplasmic twist expression is detected in rare cells within intestinal crypts in LOI( ) mice. (F) twist is greatly increased within intestinal crypts of LOI(+) mice. Scale bars, 10µm. Fig. S6. In situ hybridization analysis of Igf2 mrna levels in mouse gut with mutation in the H19 DMR (142* mouse). The composite bright- and darkfield images represent: (A) Fetal (E16.5) gut in a 142* x SD7 cross, antisense Igf2 riboprobe. (B) Fetal gut, 142* x SD7 cross, sense probe. (C) Adult (153 day) gut, 142* x SD7 cross, antisense probe. (D) Fetal gut, SD7 x 142* cross, antisense probe. (E) Fetal gut, SD7 x 142* cross, sense probe. (F) Adult gut, SD7 x 142* cross, antisense probe. Fig. S7. In situ hybridization analysis of H19 mrna levels in E16.5 mouse embryos with mutation in the H19 DMR. (A) Brightfield view over the gut in a 142* x SD7 fetus using antisense Igf2 riboprobe. (B) Darkfield view, 142* x SD7 fetus, antisense probe. (C) Brightfield view, SD7 x 142* fetus, antisense probe. (D) Darkfield view, SD7 x 142*
Sakatani et al. 16 fetus, antisense probe. (E) Brightfield view, SD7 x 142*, sense probe. (F) Darkfield view, SD7 x 142* fetus, sense probe. Fig. S8. Musashi1 immunostaining of normal colon of a colonoscopy patient without LOI and a patient with LOI. (A) musashi1 positive cells were rarely observed in colonic crypts of patients without LOI, and there was no surface staining. A higher power view of the crypt indicated by an asterisk is available in Fig. 2E. (B) In contrast, aberrant musashi1 protein expression was detected in patients with LOI throughout colonic crypts with extension to surface epithelium (surface indicated by arrow). A higher power view of the crypt is available in Fig. 2F. Scale bars, 10µm.
Sakatani et al. 17 Table S1. Semi-quantitative analysis of musashi1 staining in intestinal crypts. The number of Mushashi1-positive cells was analyzed in LOI( ) Min mice and LOI(+) Min mice, and the number of crypts containing 6 and <6 Musashi1 positive cells is shown. P value was calculated by Fisher exact test. Genotypes 6 musashi1(+) cells Number of crypts <6 musashi1(+) cells P value LOI( ) Min 5 35 LOI(+) Min 17 23 <0.01
Sakatani et al. 18 Supporting reference list S1. P. P. Zumstein, C. Luthi, R. E. Humbel, Proc. Natl. Acad. Sci. U. S. A. 82, 3169 (1985). S2. K. J. Valenzano, E. Heath-Monnig, S. E. Tollefsen, M. Lake, P. Lobel, J. Biol. Chem. 272, 4804 (1997). S3. N. Hizuka et al., J. Clin. Endocrinol. Metab. 83, 2875 (1998). S4. J. Blahovec et al., J. Endocrinol. 169, 563 (2001).